Professional Documents
Culture Documents
AN EXPERIMENTAL
STUDY OF JET NOISE
PART I: TURBULENT
MIXING NOISE
H. K. TANNA
Lockheed-Georgia Company, Marietta, Georgia 30063, U.S.A.
(Received 10 March 1976, and in revised form 19 October 1976)
The characteristics, both spectral and directivity, of turbulent mixing noise in the far field
from subsonic and fully-expanded supersonicjet flows have been studied experimentally over
an extensive envelope of jet operating conditions (jet exit velocity and temperature). The
measurements were conducted in an anechoic room which provides a free-field environment.
The results are presented in a systematic manner, and the observed trends and dependencies
are discussed in detail. In particular, the changes in detailed jet noise features with varying
velocity and exhaust temperature are assessed independently. Empirical prediction schemes
or comparisons with recent theoretical investigations are not attempted here. However, the
isothermal jet noise results are compared with those predicted by the freely-convecting
quadrupole theories (that is, in the absence of any mean flow shrouding effects). The
discrepancies between this model and the measurements, many of which have been recently
shown to occur due to the presence of mean velocity and temperature gradients surrounding
the sources, are obtained accurately over all jet operating conditions of interest.
1. INTRODUCTION
The characteristics of turbulent mixing noise from unheatedjets operated at subsonic exhaust
velocities were examined experimentally
in great detail by Lush [l]. The inadequacies
of the
freely-convecting
quadrupole theories [2-4], which do not account for the changes in radiation
efficiency due to the shrouding of the sound sources in a jet flow by the mean flow environment, were highlighted over this restricted regime of jet operating conditions. During the last
three years, various investigations, a few of which have been experimental but most theoretical,
have shown that many ofthe observed deviations from this theoretical model can be explained,
at least qualitatively, by immersing the sound sources in a realistic mean velocity environment
that is always present in a jet exhaust.
In order to study these flow-acoustic interaction effects quantitatively
at all jet operating
conditions of interest, it is vital to obtain turbulent mixing noise data of high accuracy over
extensive ranges of jet velocity and exhaust temperature. As part of a fundamental
research
program on the generation and radiation of supersonic hot jet noise, these results are now
available, and they are presented in this paper.
In the experimental program designed to obtain this data, the characteristics,
both spectral
and directivity, of the sound field of supersonic shock-free jets were studied by measuring the
turbulent
mixing noise in the far field from four 5.08 cm diameter nozzles: namely, a
convergent nozzle for pressure ratios up to critical and three convergent-divergent
nozzles
having nominal design Mach numbers of 1.4, l-7 and 2.0, respectively. In order to avoid the
contamination
of data by shock-associated
noise, these latter were operated only at their
design pressure ratios (up to 7.4). The measurements
were conducted in a carefully designed
large anechoic room which provides a free-field environment.
The facility, test procedure and
the calibration tests conducted prior to the experimental program are described in detail in
references [5] and [6].
405
406
H. K. TANNA
The ranges of various parameters defining the jet operating conditions are as follows :
pressure ratio, P,/P, = 1 to 7.4;
stagnation or total temperature ratio, TR/To = 1 to 4.5;
jet exit velocity ratio, VJ/aO= 0.3 to 3.0;
static temperature ratio, TJ/To = 0.5 to 4.0;
jet density ratio, pJ/po = 0.25 to 2.0.
Results from 75 jet exit conditions (T,/T,,
VJ/ao)
are presented here. The test program,
described in section 2, was carefully planned in order to obtain results (i) at constant jet
efflux temperature with varying jet velocity, and conversely (ii) at constant exit velocity with
varying jet efflux temperature.
The results at 90 to the jet axis were presented by Tanna et al. [5], and the effects of flow
temperature on jet mixing noise equivalent source strengths also were identified and
quantified in terms of a simple semi-empirical scaling law model to an accuracy of 1 or 2 dB.
The corresponding results at all angles to the downstream jet axis (15 < 0 6 135) are presented here. The data from the entire test program are tabulated in reference [7] for future use.
The data are analyzed and presented in a suitable form so that detailed comparisons with
the on-going theoretical efforts can be carried out at appropriate stages. The velocity dependence and directivity, both for overall results and l/3 octave band results at fixed values of
source Strouhal number, are displayed in a systematic manner. The characteristics of turbulent mixing noise from fully-expanded jet flows, as one proceeds from the low subsonic jet
velocities to high supersonic velocities, are established in detail, and the changes in jet noise
features with varying velocity and exhaust temperature are assessed independently.
Every effort is made to keep the majority of the figures self-explanatory, and the corresponding descriptions therefore are kept to a minimum. It is also pointed out that no attempt is
made to formulate a general empirical prediction method from these results at this stage,
although it is felt that this can be achieved very successfully by spending the required effort
for it.
The basic aim of this paper is to present and describe the major experimentally observed
features of jet noise, and hence comparisons with recent theoretical investigations are not
included. Some of these comparisons have been published by Tester and Morfey in a separate
= 1) are compaper [8]. However, the results for isothermaZ jet operating conditions (T,/T,
pared with the Lighthill/Ffowcs Williams/Ribner [2-4] theory of quadrupoles convected in
free space in order to (i) obtain as much insight as possible from use of this simpler model,
(ii) highlight the inadequacies of this model quantitatively, and (iii) define the regimes over
which the discrepancies occur. In addition, these comparisons provide a convenient method
of data presentation.
For each aspect under examination (e.g., velocity dependence, directivity, etc.) the data
are essentially presented in two parts. The first part describes the variation with jet exit velocity
ratio, VJa,, and hence deals with the results for which TJ/Tois unity, whereas the second part
considers the variation with jet efflux temperature ratio, T,/T,,,
at fixed jet velocities.
2. EXPERIMENTAL PROGRAM
The experimental program, summarized earlier, may now be described in detail with the
help of the experimental program chart shown in Figure 1. The chart essentially shows the
envelope of the jet static temperature ratio, TJTo (and hence also jet density ratio, pJp,),
and jet exit velocity ratio, V,/a,,, as a function of stagnation temperature ratio T,/T,,
and
pressure ratio P,/P,.
The lower limit of VJ/ao is set by the measuring instrumentation noise, as
described in reference [5]. The four nozzles employed are the convergent nozzle and the
M = 1.4, I.7 and 2-O convergent-divergent nozzles. In this study of turbulent mixing noise,
TURBULENT
407
MIXING NOISE
5.0[
IO.20
0.2
0.3
0.4
0.5
0.6
0.7
o+lo.9
I.0
I.5
2.0
2.5
3.0
4.0
J%
3. RESULTS
FROM
FREELY-CONVECTING
QUADRUPOLE
THEORY
In view of the comments made in section 1, the results from the freely-convecting
quadrupole
theories which will be used in the data presentation
are quoted in this section.
The far-field acoustic intensity, Z, is givea by the Lighthill/Ffowcs
Williams/Ribner
theory
for quadrupole
sources convected in the absence of any mean velocity or temperature
gradients, in dimensional
form, as
D(e)
al R2 [( 1 - M, cos 0)2 + LX~
Mf(cos 8 + e2 sin2 0)]5/2
(1)
408
H. K. TANNA
where pm = density in the mixing region, p,, = density of the external fluid, a,, = ambient
speed of sound, VJ = jet exit velocity, D = nozzle diameter, R = distance from the jet to the
observer, 8 = angle between the direction of emission of sound and the downstream jet axis,
D(0) = directivity of the unconvected quadrupole distribution in the jet, A4, = eddy convection Mach number (i.e., quadrupole convection speed divided by the ambient speed of sound,
VJa,), a = turbulence parameter expressing the rate of decay of the turbulent eddies, given
by 01,/l,, where w is the typical frequency of turbulence observed in a reference frame
moving at the eddy convection speed and I, is the longitudinal length scale of the turbulence,
and e = ratio of lateral to longitudinal scales of turbulence I,//,.
In equation (1) the amplification due to source convection at high velocities is given by the
factor
[(l - MCCOS 6) + 2 M:(cosZ 6 + e2 sin2 8)]-52,
(2)
which is commonly referred to as the convective amplification factor.
Since the data correlation at 90 to the jet axis for isothermal jets reported previously was
better when based on a V;1.5dependence rather than the normally accepted velocity index of 8
(see reference [5]), all comparisons presented here will be based on the V;1.5dependence.
For isothermal jet operating conditions (i.e., pm/p,, = l), and upon assuming the basic
unconvected quadrupole distribution in the jet to be statistically isotropic (i.e., spherical
directivity, D(O) = l), and putting e = 1 for convenience, equation (1) for the overall intensity
therefore gives (a) velocity dependence at angle 0 as
( V&-J5[( 1 - MC cos ey + a2 My2
(3)
and (b) directivity at fixed VJ/ao as
[( 1 - M, cos l9)+ c? A4:1-5/2.
(4)
For the analysis of the results in l/3 octave bands, the same expressions (3) and (4) above can
be used for the velocity dependence and the directivity, except that, in this case, since the
sources are being convected at an eddy convection Mach number, MC,the observed frequency
_fand the source frequency& will be related by the Doppler factor, F, according to
f, =fF =f[(l - MCcos 0) + t12A4:]2.
(5)
Thus, if the l/3 octave results are analyzed at constant source Strouhal number,
A D/V, = (fD/VJ)[( 1 - A4,cos t?)2+ C?A4:]1/2 = constant,
(6)
then the l/3 octave intensities will be independent of frequency, as explained in detail by Lush
[l]. In the analysis of the results at constant source Strouhal number, the observed frequency,
f, corresponding to the fixed source frequency,f,, is first determined from equation (5), and
the Strouhal number correction ensures that the intensity is measured at the same point on
the spectrum relative to the peak. In this manner, the source in a geometrically similar position
in the jet is observed, irrespective of the emission angle, jet diameter and jet velocity.
In the following sections, the velocity dependence and the directivity of both the overall
intensity and the l/3 octave intensity are studied in the manner described briefly above. In
comparing the results with the freely-convecting quadrupole theory, MCis taken to be 0.67 VJ/aO
and tl = 0.3, as measured for low velocity jets by Davies et al. [9].
Note: throughout the rest of this paper, the freely-convecting quadrupole theory will be
referred to as theory for simplicity. It should not be confused with any recent theoretical
efforts at either Lockheed-Georgia or elsewhere.
4. VELOCITY DEPENDENCE
4.1.
ISOTHERMAL
(T#,,
OF OVERALL INTENSITY
1) JET
The measured overall intensity as a function of V,/a, at four angles to the jet axis, namely
135, 90, 45 and 22.5, is shown in Figure 2, together with the theoretical predictions for
409
140.
I 30.
z
0
a
C
2
u
.E
ZZ
e
120.
I IO.
100.
F
0
90.
80.
70.
601
0.3
0.4
0.5
0.6
0.7
0.60.9
1.5
I.0
2.0
2.5
3.0
vJ/aO
velocity dependence given by equation (3). The theoretical curves at 8 = 135, 45 and 22.5
are plotted relative to the theoretical curve at 90 and therefore show the correct amount of
convective amplification. At f3= 90, the theoretical velocity scaling follows the measured
velocity dependence very accurately over the entire subsonic and supersonic velocity range
considered here. However, in the forward arc (0 > go), the theory underestimates the measurements, and at f3= 45 and 22.5, the theory grossly overestimates the measured velocity
dependence. The amount of overestimation in the rear arc (0 < 90) increases as t3decreases
or as VJ/ao increases.
It should be noted that although the theoretical curves in Figure 2 can be shifted much closer
to the measurements by choosing appropriate values of M, and c1(e.g., MC = O-5V,/a, and
a = 0.55 as suggested by Ribner), there is no experimental evidence to justify this at the
present time.
601
0.3
0.4
0.5
0.6
0.7
0.60.9
I.0
I.5
2.0
2.5
b/a0
Figure 3. Effect of TJ/rO on velocity dependence of overall intensity: B = 90. T,/T, (nominal):
o, 1.0; ?,?2.3; o, 3.4.
2, cold;
410
H. K. TANNA
The most general effect of heating can be observed by examining the velocity dependence
of overall intensity at 19= 90 for various values of TJT,,. This is presented in Figure 3. As
noted first by Hoch et al. [IO], and as commonly accepted now, the effect of heating is to
produce an increase in noise at low jet velocities, whereas at high jet velocities the overall
level decreases with heating. The velocity exponent decreases with heating, and from the
figure it appears that the value of VJ/ao at which the effect of heating (relative to TJ/To = 1)
reverses increases as TJfTo increases.
It should be noted that no theoretical velocity dependence is implied in this figure. The
full lines shown are simply lines joining experimental points. This comment in fact applies to
all subsequent figures where the variation with TJ/To is presented.
5. DIRECTIVITY OF OVERALL INTENSITY
5.1. ISOTHERMAL (T,/T, = 1) JET
The directivities of measured overall intensity at several values of VJ/ao from 0.5 to I.95
are shown in Figure 4. At subsonic velocities, the intensity is directional with a peak quite
close (20-25) to the jet axis, and the effect of refraction is to produce a slight dip at angles
less than approximately 20-25. At supersonic velocities, the overall intensity is highly
directional; the peak angle and the dip at angles smaller than the peak angle increase progressively as thejet velocity increases. It should be noted, however, that at these high velocities,
these dips at small angles to the jet axis do not arise purely due to refraction, but are a result
140
I
130.
z
9
2.
c
c
u
.t
1
e
z
0
120.
HO-
100.
90.
80
701
0
1
I5
30
Directlvity
1
45
I
60
angle
v
A
V
A
I
75
from
1
90
I
105
downstream
T,/T,*
1. -,
8
120
jet
axis,
I50
8 (degrees)
W0
TP
0.50
8
12
51
63
0.90
1.33
1.95
b
135
theory.
TURBULENT
MIXING
411
NOISE
of the combination
of refraction and Mach wave radiation [3] that occurs when the eddy
convection Mach number in the direction of the observer approaches unity (i.e., M,cos B fi 1).
Further work is required to establish which of these two phenomena
is dominant
in this
velocity regime.
The measured directivities are now compared with the theoretical directivities given by
equation (4), with IV, = 0.67 V,/a,, and c1= 0.3. It should be noted that all theoretical directivities are plotted relative to the experimental
value at 0 = 90, and not adjusted vertically for
best fit; thus they represent true comparisons between theory and experiment. The theoretical
convective amplification in the rear arc and convective attenuation
in the forward arc are
clearly too large. The deviation increases progressively as VJ/aOincreases and also as 0 increases
or decreases from 90. Even at the lowest jet velocity considered ( VJ/ao = 0*5), the agreement
between theory and experiment is poor.
Again, as in the case of velocity dependence (section 4.1), the theoretical convective amplification can be reduced significantly by selecting appropriate
values of M, and c(, but this
cannot be justified until measurements
of M, and CIover the entire velocity range become
available.
5.2.
EFFECT OF
90
70
0
15
30
45
60
Directivity
angle
75
from
90
105
downstream
jet
120
axis,
135
150
8 (degrees)
TP
a
Fi
0.95
23
1.0
3
32
8
3.4
68
412
H. K. TANNA
Directivity
A
0
0
D
Directivity
T,/To (nominal)
TP
0.84
1.0
2.3
3.4
7
12
36
72
TP
0.57
62
:
D
23
1.2
3.4
52
39
75
6. l/3 OCTAVE
VJ/uo= 0.9.
V&&I= 1.47.
SPECTRA
413
= 1) JET
The l/3 octave spectra at various jet exit velocities in the range from VJ/ao = 0.5 to 1.95
are shown at 0 = 90 and 30 in Figures 8 and 9, respectively. At 0 = 90, the spectra are
broad and smooth with peak frequencies scaling on the Strouhal number basis. However,
at 30 to the jet axis, the spectrum shape changes and the peak becomes more marked and
shifts to a lower frequency. Furthermore,
the peak frequency does not scale as a Strouhal
number and remains essentially constant at about 1400 Hz over the subsonic velocity range.
In examining the corresponding
results at all angles to the jet axis (not shown here) between
90 and 15, it was observed that the variation in spectral shapes at the supersonic velocities as
one proceeds from 90 to 15 was rather interesting. At 8 = 90, the spectra throughout the
velocity range are parallel in nature. This parallel stacking
is retained as the observer
angle 0 reduces to approximately
40. It suggests that for 0 > 45 the eddy convection Mach
number in the direction of the observer, MC cos 0, is less than unity at all values of VJ/aO up to
2.0, and therefore the convective amplification
increases steadily as 0 reduces from 90 to
45. At smaller (0 < 45) angles, however, it was observed that while the spectra at all subsonic
velocities and the highest supersonic velocity considered (V,/a,, = 1.95) essentially exhibit
the parallel stacking pattern, the spectra for VJjao = 1.33 and l-65 do not conform to this
pattern. At these two velocities, the spectrum levels are higher, especially at the higher
frequencies, and the spectra therefore peak at frequencies which are higher than the peak
frequencies for the spectra at all other velocities considered. The magnitudes of these observed
effects increase as the observer moves further towards the jet axis, and at 0 = 30 (Figure 9),
the spectrum for VJ/ao = 1.65 is in fact identical in level to the spectrum for VJ/aO = 1.95 for
120
60
.
.
.
t
501
0.25
0.5
173
Octave
center
4
frequency
16
31.5
(kHz)
Figure 8. Variation of l/3 octave spectra at 0 = 90 with jet velocity: T./T, e 1. TP: v, 8; o, 10; A, 12;
v, 51; .,58;
n, 63.
414
H. K. TANNA
130
I
Va0
n
a
120-
D .
n.
* 1.65
rJov
Q -1.33
. .
A .
I
.
I
.
0
0
0
0.5
frequency
(kl-iz)
60 -.
-0
16
._
31.5
Figure 9. Variation of l/3 octave spectra at 0 = 30 with jet velocity: T,/T, A 1. TP: v, 8; 0, 10; A, 12;
o,51;.,58;n,63.
T,jT,
TURBULENT
l/3
Octave
415
MIXING NOISE
center
frequency
CkHz)
Figure 10. Effect of rJ/TO on l/3 octave spectra at low jet velocity: 0 = 45, V&z,,= 0.5.
IO0
TJ/T (nominal)
TP
a
:
0.95
2.3
1.0
3
32
8
3.4
68
l/3 Octave
center
frequency
(ktiz)
Figure 11. Effect of rJ/TO on l/3 octave spectra at medium jet velocity: 0 = 45, V,/ao = 0.9.
TJ/To (nominal)
TP
no
0.84
2.3
1.0
7
36
12
3.4
72
these frequencies in fact decrease with jet heating. In contrast, the effect at higher frequencies
does not reverse as one proceeds from low to high VJ/ao; the reductions in spectrum levels at
these higher frequencies increase in magnitude as VJ/aO increases throughout the speed range.
These observations
are in agreement with the results of Hoch et al. [lo].
At all velocities considered, the effects of jet heating described above alter the peak fre-
H. K. TANNA
416
80
0.25
o-5
l/3 Octave
center
frequency
16
31.5
(kHz)
Figure 12. Effect of T,/T, on l/3 octave spectra at high jet velocity: 0 = 45, VJ& = 1.47.
T,/T,
(nominal)
TP
0.57
1.2
2.3
3.4
62
52
39
75
n
0
n
V
quencies in a consistent manner; the peak frequency decreases as the jet exit temperature ratio
increases from unity. This suggests that it may be possible to scale the peak frequencies as a
function of temperature,
and this is discussed in the following subsection.
6.3.
PEAK FREQUENCY
8.0.
6.3.
5.0.
4.0.
3.15.
s
a!
2.5.
2.0.
~~2~L__$._i~jo
f-
0.3
0.4
0.5
0.6
0.7
0.80.9
I.0
I.5
vJiaO
TURBULENT
417
MIXING NOISE
off, D/V, = l-5, compared to the Strouhal number of 0.8 obtained by Lush [I] for data from
an unheated jet over the subsonic velocity range. In contrast, at 0 = 22*5, the peak frequency
does not scale on the Strouhal number basis. For V,/a,, < 1, the peak frequencies remain
constant at approximately
1100 Hz, which corresponds to the Helmholtz number off, D/a0 =
O-16 and therefore agrees with f,D/u0 = 0.15 obtained by Lush. As the jet velocity increases
above V,/u, = 1, the peak frequency at 8 = 22.5 first increases, as noted previously, and
then decreases as the velocity is increased beyond V,/a,, ti 1.5. For V,/u, > 2, the peak
frequencies would presumably again scale on f,D/u, e O-16.
I.8 r
0.5
I.0
I.5
2.0
2.5
3.0
3.5
TJ/To
Figure 14. Variation of peak frequency with jet velocity and temperature:
?,0.6;
? 0,0.7; CJ,0.8; f, 0.9; x, 1.16; 0, 1.33; A, 1.47; W, 1.65; +, 1.80; T, 2.00; +, 2.30; 0, 2.55.
The variation of peak frequency with jet temperature was examined in detail at 8 = 90,
and the results are presented in Figure 14. This figure represents the peak frequencies obtained
from l/3 octave spectra for approximately
70 jet operating conditions, covering the range of
V,/u,, from 0.4 to 2-55 and the range of TJT,, from 0.57 to 3.4. At each nominal value of
TJ/To considered, the peak frequencies scale on the Strouhal number basis, as shown by the
clusters of experimental points in Figure 14. The variation of the peak Strouhal number with
temperature is shown by drawing a smooth curve through the points and the result is rather
interesting. It shows that the peak Strouhal number is maximum when the jet exit temperature
is close to the ambient temperature,
and it decreases progressively as TJ/To either increases
or decreases from unity.
7. VELOCITY
DEPENDENCE
CONSTANT
SOURCE
7.1.
ISOTHERMAL
AT
(TJT,, = 1) JET
The turbulent mixing noise data for isothermal jet operating conditions over the range of
V,/u, from 0.5 to 1.95 have been analyzed in l/3 octave bands at particular values of the
source Strouhal number fsD/ VJ (equation (6)), in accordance with the technique adopted by
Lush and which was described briefly in section 3. Here again one must bear in mind that in
all analyses presented in this paper MC is taken to be equal to O*67VJ/u0 and CIto 0.3, as
suggested by jet flow measurements
at low velocities [9].
The dependencies
of l/3 octave intensities on jet velocity at three angles to the jet axis,
namely 90, 45 and 22.5, are shown in Figures 15, 16 and 17, respectively. The five values of
418
H.
K.TANNA
120
to-
50.
40
0.3
0.4
1
0.5 0.6 0.7 0.80.9 I.0
I.5
2.0
2.5 3.0
WO
Figure 15. Velocity dependence of l/3 octave intensity at 0 = 90: T,/TO i I._&D/ VJ: o,
Freely convecting quadrupole theory.
a, 3.0. -)
130
120
110
G
D
>. 100
.Z
E
2
90
z
0
g
80
9
70
60
501
0.3
0.4
I
0.5 0.6 0.7 0.80.9 I.0
I.5
2.0
2.5
vJ Ia0
Figure 16. Velocity dependence of l/3 octave intensity at 0 = 45: T,/To t 1. h D/ VJ: A, 0.03;
V, @3; 0, 1.0; a, 3.0. -,
Freely convecting quadrupole theory; ----,
experiment.
0,
0.1;
TURBULENT
419
MIXING NOISE
130,
120.
60 .
501
0.3
0.4
O-5
0.6
0.7
0.80.9
I.0
I.5
2-O
2.5
VJ 100
Figure 17. Velocity dependence of l/3 octave intensity at 0 = 22.5: T,/T,G 1. fs D/ V,: 0, 0.1;
experiment.
0, 1.0; A, 3.0. -,
Freely convecting quadrupole theory; ----,
TJ,
0.3;
variation,
the magnitude
of the overestimation
increasing progressively
with increasing
frequency.
The velocity dependence at 8 = 22.5 (Figure 17) shows similar features, except that the
measurements
begin to fall below the theoretical velocity dependence at a lower value of the
frequency parameter:
i.e., fS D/V, = 0.1. At frequencies
below this value, the predicted
variation of intensity with jet velocity (solid line) is approximately
in agreement with measure(broken line) tend to
ments, whereas at frequencies above f,D/VJ = 0.1, the measurements
fall well below the expected variation.
7.2.
EFFECT
OF JET TEMPERATURE
T,jT,
The influence of temperature on the velocity dependence of l/3 octave intensity is illustrated
at 90 to the jet axis. The two extremes of the frequency parameter, namely f,D/ VJ= 0.1and
3.0, are considered in Figures 18 and 19, respectively. As in section 4.2, no theoretical velocity
40.1
0.3
0.4
0.5
o-5
0.7
0.80.9
I
I.0
I.5
2-O
2.5
I
3.0
L/J/a0
420
H.
K. TANNA
scaling is implied in these figures. The curves through the experimental points are shown
merely to illustrate the measured trends.
At low source Strouhal number (fs D/V, = O-1, Figure 18), the effects of jet heating are
qualitatively similar to the effects observed for the velocity dependence of overall intensity,
described in section 4.2: that is, the noise increases at low velocities, and the noise decreases
at high velocities. The velocity exponent decreases at elevated temperatures. It also appears
that the break-point (value of V,/a, at which the effect of heating relative to T,/T,, = 1 reverses)
increases as T,IT, increases.
501
0.3
0.4
I
0.5 0.6 0.7 0.60.9 I.0
I.5
2.0
2.5 3.0
At high source Strouhal number (fs D/ VJ = 3.0, Figure 19), this break-point is shifted to a
much lower value of VJ/uo. Reductions in noise levels are now measured almost throughout
the velocity range considered, and the magnitudes of these reductions with heating increase
significantly as VJ/ao increases.
AT CONSTANT
8. DIRECTIVITY OF l/3 OCTAVE INTENSITY
STROUHAL NUMBERS
SOURCE
ISOTHERMAL
(T,/T, = 1) JET
The directivities at various jet exit velocities for the same five values offs D/ VJ considered
previously are shown systematically in Figures 20 through 24. The directivities due to quadrupole convection in free space (equation (4)) are also shown as solid lines, and it should be
noted that since they are plotted relative to the experimental values at 0 = 90, and not
421
Y4\
l20-
m
s
IIO-
a.
C
z
f
IOO2
8
9
90
80-
70
0
15
Directivlty
30
45
angle
from
60
75
downstream
I05
90
jet
axis,
I20
135
@(degrees)
Figure 20. Directivity of l/3 octave intensity ath D/VJ= 0.03: TJ/TOs 1. -,
pole theory; ----,
same theory with modified 0 (source at X/D = 10); ----,
loi
I20
Directivity
angle
from
downstream
jet
axis,
I/J/so
TP
I.95
63
I.33
51
8 (degrees)
adjusted vertically to give the best fit, they represent the correct amount of convective amplification (or convective attenuation as the case may be) relative to the level at 90.
At the two lowest values of the non-dimensional frequency parameter, namelyf, D/ VJ =
0.03 and 0.1 (Figures 20 and 21), the theoretical directivities are in fairly good agreement with
measurements, except at small and large angles to the jet axis, where the measured levels are
consistently higher. The amount of this underestimation at low frequencies tends to increase
as the jet velocity increases. At angles between approximately 45 and 90, the agreement
between theory and experiment can be improved significantly by using the following argument. At these lower frequencies, recent source location studies indicate that the sources
422
H. K. TANNA
-1
501
0
15
30
45
Directivity
.
75
60
angle
from
90
,
105
downstream
120
jet
axis,
,
I35
I.95
63
64
1.33
51
58O
0.9
12
510
.
I50
L?(degrees)
Figure 22. Directivity of l/3 octave intensity ath D/ V, = 0.3 : T.JTo* 1.--,
Freely convecting quadrupole
theory; --- -, same theory with modified 0 (source at X/D = 10); ----,
experiment.
;A,
_i
o,5 8 420
601
0
15
30
Directivity
45
60
75
angle from
90
105
downstream
120
jet
135
150
axis, 8 (degrees)
in the jet flow are located at some distance downstream of the nozzle exit plane. The center of
the observer arc should therefore be taken in the region where the sources exist, rather than
at the nozzle exit plane. In the present comparison, if the sources at these lower frequencies
are assumed to be located at approximately 10 nozzle diameters downstream, which appears
to be reasonable from available evidence, then the observer angles are modified and the
corresponding modified theoretical directivities are also shown in Figures 20 and 21. It can
be seen that while this procedure makes little difference at the extremes of observer angle
TURBULENT
423
MIXING NOISE
Ii0
7
60 -
50
0
,5
30
45
60
75
90
105
120
135
150
Direc+ivi+y
anglefrom downstream jet axis, 0 (degrees)
Figure 24. Directivity of l/3 octave intensity at&O/ VJ = 3.0: TJ/TOt 1. (a) -,
Freely convecting quadrupole theory (5 powers of Doppler factor); (b) ----,
3 powers of Doppler factor; --,
experiment.
(except at the highest velocity, where M,cosB becomes greater than unity at small angles),
the agreement between theory and experiment at 6between 45 and 90 is considerably
improved at all jet velocities.
It is admitted that the above procedure does introduce a slight deterioration
in the correlation at some angles in the forward arc, but in general a careful examination
reveals that the
amount of benefit gained with this procedure outweighs the slight damage. Besides, it is
physically correct to use this procedure at the lower frequencies.
At the next value of source Strouhal number, fSD/VJ = 0.3 (Figure 22), the agreement
between theory and experiment is again reasonable between 9 ti 45 and 90, and this can be
improved significantly by using the above argument of choosing the correct source location.
At smaller angles to thejet axis, however, the theory now overestimates the l/3 octave intensity,
showing that refractive effects are beginning to emerge at this frequency. The overestimation
in fact becomes worse as the jet velocity increases. Furthermore,
in Figure 22, it is interesting
to note the observation, made previously for the directivity of overall intensity, that while the
directivities for VJ/ao = O-5,0*9 and 1.95 are parallel in nature, the directivity for V,/ao = 1.33
does not fall into this pattern. The levels at smaller angles are higher for VJ/aO = 1.33.
Whether this is due to MCcos B approaching unity or the consequence ofthe refractive phenomena needs to be ascertained.
At the next higher frequency given by f,D/VJ = 1.0, Figure 23 shows that the effects of
refraction are present even at larger angles to the jet axis. The freely-convecting
quadrupole
theory (directivity denoted by (a) in Figure 23) now overestimates in the rear arc (0 < 90C)
and underestimates
in the forward arc (0 > 90) by large amounts at this frequency. The
magnitude
of the discrepancy in general increases as the jet velocity increases or as the
observer moves away from the 90 pivot. Also shown in Figure 23 are theoretical directivities
based on three powers of Doppler factor (denoted by (b)), i.e.,
z - [(l - M,
cos
ey
k.ff]-32
with
and
CI= 0.3,
(7)
as opposed to the normally accepted five powers for freely convected quadrupoles.
It can be
seen that at angles greater than the cone of silence angle, given by 8, = set-( 1 + M,) from
ray acoustics, the directivities predicted by three powers of Doppler factor are in very good
H. K. TANNA
424
agreement with the measurements, except at the highest velocity, VJ/uo = l-95, where some
overestimation still occurs. This observed angular dependence of intensity at high frequencies
and at angles outside the zone of silence is iti agreement with the results of the on-going
theoretical work by Tester and Morfey [S].
Finally, for the comparisons at the highest frequency given by f, Dl V, = 3-Oand shown in
Figure 24, the above comments for Figure 23 apply here as well. In addition, it can be seen
that since the frequency is higher in this case, the effects of refraction are even greater than
those observed in Figure 23. The theoretical directivity given by equation (4) does not
resemble the measured directivity shape at all. In contrast, the angular dependence given by
equation (7) is in very good agreement with the measurements roughly outside the zone of
silence (0 > BC,see Figure 24).
8.2.
TJ/TO
The influence of jet exhaust temperature on the directivity of l/3 octave intensity was
investigated in detail, and the purpose ofthis subsection is to present the major effects throughout the velocity range in sufficient detail. Specifically, the variation of directivity with temperature is presented at the two extremes of the source Strouhal number for three jet efflux
velocities, V,/a, = 0.5, 0.9 and l-47. The results are shown in Figures 25 through 30. The
curves through the experimental points in these figures are provided purely in order to illustrate the measured trends; they do not imply any theoretical scaling of angular dependence
with temperature.
8.2.1. Low frequency
At low source Strouhal number,f, D/VJ = O-1, the effect of TJ/TO on the directivity as
proceeds from low V,/a, to high VJ/ao is most revealing-see Figures 25,26 and 27. At
velocity ( VJ/aO= 0.5, Figure 25), the intensity at all angles increases with heating, and
magnitude of this increase is nearly independent of 8; the directivities at various TJT,,
go:
80 -
z
9
>I
a
;
.-
70 *
TJ/%
3.4
2.3
60.
8
I.0
P
50 -
40
0
15
30
45
60
Directivity
75
angle
from
90
downstream
105
120
135
150
17
0.95
1.0
2.3
3.4
TP
3
8
32
68
one
low
the
are
TURBULENT
MIXING
NOISE
425
110
100. J
I.0
z
0
z.
E?
2
.E
z
0
3.4
90
80.
s
T
-
G / r,
3.4
70 -
I.0
60,
0
15
30
45
60
Directivity
75
90
I05
120
135
150
Figure 26. Effect of T,/T, on directivity of l/3 octave intensity at V&z,, = @9:f, D/ V, = 0.1.
TJ/To (nominal)
TP
0.84
1.0
2.3
3.4
7
12
36
72
A
0
0
V
130
I.2
P3,3.4
80
0
I5
30
45
60
Directivity
75
90
105
120
135
150
Figure 27. Effect of T,/T, on directivity of l/3 octave intensity at V&z,,= 1.47: fs D/V, = @ 1.
A
0
cl
V
TJ/To (nominal)
TP
0.57
1.2
2.3
3.4
62
52
39
75
virtually parallel. In contrast, at high velocity ( VJ/aO= 1.47, Figure 27), the intensity decreases
with heating. However, in this case, the magnitude of this reduction is indeed angular dependent. At medium values of jet exit velocity ( VJ/ao = 0.9, Figure 26), the effect of T,/T,is not
so dramatic, as observed previously for the overall intensity results. There is a slight increase
in level at most angles, but a small reduction near the downstream jet axis.
426
H. K. TANNA
80
9cfi
70.
h
.t
f
.E
60-
rJ/
I.0
2.3
0
1
50
3.4
40
0
15
30
Directivity
45
60
angle from
75
90
105
downstream
120
135
I i0
Figure 28. Effect of TJ/To on directivity of l/3 octave intensity at VJ/ao= 0.5: f, D/ V, = 3.0.
T,/To (nominal)
TP
0,
0.95
I.0
2.3
3.4
3
8
32
68
41
42
54
60
n
0
??
100,
2.3
3.4
501
0
15
30
45
Directivity
60
75
90
105
120
135
150
Figure 29. Effect of TJ/To on directivity of l/3 octave intensity at VJ/uO= @9:f, D/k, = 3.0.
n
0
cl
V
8.2.2.
TJ/To (nominal)
TP
0,
0.84
1.0
23
3.4
7
12
36
72
50
51
60
64
High frequency
At high source Strouhal numbers, the effect of heating remains consistent throughout
the
velocity range, as can be seen in Figures 28,29 and 30: that is, the noise levels decrease with
increasing T,/T, at all velocities. In addition, these noise reductions are angular dependent
at all velocities, especially at smaller angles to the jet axis. For a fixed value of V,/u,, the peak
radiation angle, 8,, increases as the static temperature
ratio increases. At observer angles
smaller than 8, ,the intensity falls off rapidly, and this effect is enhanced with increasing T,/T,.
TURBULENT
80
MIXING
421
NOISE
3.4
0
I5
30
45
60
Directivity
75
angle
from
90
downstream
I05
120
pet axis,
I35
150
@(degrees)
Figure 30. Effect of 71J/r, on directivity of l/3 octave intensity at VJ/ao = 1.47: AD/V, = 1.0.
.-___.L
0
0
V
TJ/TO (nominal)
___._
0.57
1.2
2.3
3.4
TP
62
52
39
75
8,
~~
57
61
65
68
An attempt was made to reconcile the rapid decrease in level at small angles as a function
of T,/T, in terms of the cone of silence angle, 8,. For heated jets, this angle is given from
ray acoustics as
cos 8, = a,/(a, + V,),
(8)
where a, is the speed of sound in the source region, and is related to the source temperature T,
= (TJTo) . As obtained by using the expression for (T,/T,J discussed in reference
[5], i.e.,
TJT, = 0*7[(T,/T,.,)- 11+ 1,
(9)
by (4ao)
and putting MC = O-67(VJ/a,), values of BCfor the jet operating conditions (VJu,, TJT,,)
considered here are tabulated in Figures 28, 29 and 30 for the high frequency directivities.
At the two subsonic velocities (Figures 28 and 29), it can be seen that for every value of T,/T,,,
the intensity increases as the observer moves from the forward arc towards the downstream
jet axis, until the cone of silence angle 0, is reached. Beyond this, the levels fall rapidly as
expected from simple physical considerations. At VJ/uo = 1.47, however, this concept does
not appear to shed any light on the measured directivities, except at the highest temperature
ratio of 34---see Figure 30.
9. CONCLUDING
REMARKS
The characteristics of turbulent mixing noise from fully-expanded static jet exhausts have
been investigated experimentally in great detail over an extensive envelope of jet operating
conditions. The results, both overall and spectral, have been displayed in a systematic and
comprehensive manner as to velocity dependence and directivity. The major features throughout the ranges of jet efflux velocity and exhaust temperature have been discussed in sufficient
detail. In particular, the carefully planned experimental program has made it possible to make
an independent assessment of the effects of velocity and temperature on jet noise.
The data presentation was aided by the familiar freely-covnecting quadrupole model (i.e.,
a model in which the presence of mean velocity and temperature fields surrounding the
sources is neglected) forjet noise. The nature of the inadequacies of this model, the magnitudes
of the discrepancies involved, and the regimes over which they exist have been studied for
most aspects of data presentation.
428
H. K. TANNA
Recent theoretical investigations [8, 11, 121rigorously indicate that many of the observed
deviations between experiment and the basic freely convecting quadrupole model are a consequence of the shrouding of the sound sources in the jet flow by mean velocity and temperature
gradients. In order to obtain greater physical understanding of such phenomena and to
establish the effects of these flow-acoustic interactions quantitatively throughout the jet
operating conditions (velocity and temperature) of practical interest, the theoretical studies
need to be continued. The accurate and comprehensive data obtained from the experimental
program described in this paper should prove most valuable in such research efforts.
Upon completion of this experimental study of turbulent mixing noise from shock-free
supersonic jets, the next logical step was to examine experimentally in detail the noise
characteristics of shock-containing jet flows. Measurements of shock-associated noise over
a large range of supercritical pressure ratios and jet exhaust temperatures have also been
conducted at Lockheed-Georgia Company, and the results are presented and discussed in
a companion paper (Part II) [ 131.
ACKNOWLEDGMENTS
This study forms a part of the larger research program on the generation and radiation of
supersonic jet noise, conducted at Lockheed-Georgia Company and financed by the U.S. Air
Force Aero Propulsion Laboratory and the U.S. Department of Transportation under
Contract F33615-73-C-2032. The author wishes to express sincere acknowledgments to
Dr P. D. Dean and Mr R. H. Burrin for designing the facility and for their help in conducting
the experiments. The data acquisition and reduction facilities were operated by Mr C. B. Reid
and Mr L. V. Mazzarella.
REFERENCES
1. P. A. LUSH 1971 Journal of Fluid Mechanics 46, 477-500. Measurements of subsonic jet noise
and comparison with theory.
2. M. J. LIGHTHILL1962 Proceedings of the Royal Society A267, 147-182. Sound generated aerodynamically (The Bakerian Lecture),
3. J. E. FFOWCS WILLIAMS1963 Philosophical Transactions of the Royal Society A255, 469-503.
The noise from turbulence convected at high speed.
4. H. S. RIBNER 1964 Advances in Applied Mechanics 8, 103-182. The generation of sound by
turbulent jets.
5. H. K. TANNA, P. D. DEAN and M. J. FISHER 1975 Journal of Sound and Vibration 39, 429460.
The influence of temperature on shock-free supersonic jet noise.
6. R. H. BURRIN,P. D. DEAN and H. K. TANNA 1974 in The Generation and Radiation of Supersonic
Jet Noise, U.S. Air Force Aero Propulsion Laboratory Technical Report AFAPL-TR-74-24. A new
anechoic facility for supersonic hot jet noise research at Lockheed-Georgia.
7. H. K. TANNA, P. D. DEAN and R. H. BURRIN1976 In The Generation and Radiation of Supersonic
Jet Noise, U.S. Air Force Aero Propulsion Laboratory Technical Report AFAPL-TR-76-65.
Volume III. Turbulent mixing noise data.
8. B. J. TESTERand C. L. MORFEY1976 Journalof Soundand Vibration 46,79-103. Developments in
jet noise modelling-theoretical
predictions and comparisons with measured data.
9. P. 0. A. L. DAVIES,M. J. FISHERand M. J. BARRATT1963 JournalofFluidMechanics
15,337-367.
The characteristics of the turbulence in the mixing region of a round jet.
10. R. G. HOCH, J. P. DUPONCHEL,B. J. (&KING and W. D. BRYCE 1973 Journal of Sound and
Vibration 28,649-668. Studies of the influence of density on jet noise.
11. R. MANI 1976 Journal of Fluid Mechanics 73, 753-778. The influence of jet flow on jet noise.
Part 1. The noise of unheated jets.
12. R. MANI 1976 Journal of Fluid Mechanics 73, 779-793. The intiuence of jet flow on jet noise.
Part 2. The noise of heated jets.
13. H. K. TANNA 1977 Journal of Sound and Vibration 50, 429444. An experimental study
noise, part II: Shock-associated noise.
ofjet